The Scene
It's 3 AM in Capital One's infrastructure operations center. A payment processing system is throwing errors. Engineers are digging through logs from twelve different services, correlating timestamps manually, burning hours they don't have. Somewhere in Virginia, a transaction queue is backing up. Customers are getting declined.
This is the moment Observe was built for.
Within seconds, Observe's AI SRE has correlated telemetry across all twelve services. It identifies the root cause: a misconfigured Kubernetes pod that's throttling database connections. The AI doesn't just flag the problem - it generates a hypothesis, ranks probable causes, and points engineers directly at the fix. What used to take three hours now takes three minutes.
Observe provides a centralized and pre-correlated data layer that meaningfully organizes telemetry data from many sources at scale.- Mark Cauwels, Managing VP, Enterprise Platforms Technology, Capital One
The Problem
Modern enterprises generate an incomprehensible amount of machine data. Logs. Metrics. Traces. Event streams. Every microservice, every container, every API call - all of it produces telemetry. By 2024, large companies were generating petabytes of operational data monthly.
Traditional observability tools couldn't keep up. Splunk charged by the gigabyte - and enterprises were drowning in terabytes. Datadog's pricing made comprehensive monitoring prohibitively expensive. Companies faced an impossible choice: either sample their data and miss critical signals, or go bankrupt storing it all.
The industry had accepted a compromise that shouldn't have been necessary. Engineers were making business decisions about which logs to keep, which metrics to discard, which traces to sample. They were flying blind by design.
The Founders' Bet
In 2017, Jacob Leverich left Splunk. Jonathan Trevor left Wavefront. They'd spent years building observability tools the old way - and they knew exactly where the architecture broke down.
Their bet was radical: what if you built an observability platform on top of Snowflake's cloud data warehouse? What if, instead of proprietary storage engines and index structures, you used commodity cloud infrastructure? What if the storage layer was so cheap that you could keep everything - every log line, every metric, every trace - forever?
The economics would be inverted. Instead of charging customers by the byte to discourage data retention, you'd charge them for what they actually needed: fast answers to hard questions.
The economies of scale from Observe's commodity pricing means we can consume as much data as we want from as many sources as we want and retain it for as long as we want without worrying about cost.- Richard Marcus, Head of Information Security
The Team
Jacob Leverich
Co-Founder & CPO
Former Director of Engineering at Splunk. Eight years building the technical foundation of Observe's streaming data lake.
Jonathan Trevor
Co-Founder
Came from Wavefront, where he learned what worked - and what didn't - in real-time monitoring at scale.
Jeremy Burton
CEO
Former President of Products at Dell EMC, overseeing a $15B business. Sits on Snowflake's board. The enterprise software veteran who turned a startup into an acquisition target.
The Journey: From Stealth to Snowflake
Founded at Sutter Hill Ventures' Palo Alto office
Core architecture built on Snowflake's Data Cloud
Emerged from stealth with $32M Series A
$115M raise, Snowflake invests as strategic partner
$145M Series B, AI Investigator and APM launch
$156M Series C for AI-native expansion
Acquired by Snowflake for approximately $1B
The Product
Observe built a streaming data lake that ingests anything with a timestamp. Logs, metrics, traces - it doesn't matter. Everything goes into a single, unified data layer running on Snowflake. No separate tools. No data silos. No reconciling timestamps across different systems.
The magic happens in what they call "incremental views." Instead of pre-aggregating data and throwing away the raw signals, Observe materializes relationships in real-time. When you search for a failing request, you're not querying pre-built indexes - you're exploring the actual telemetry, with machine learning that identifies semantic relationships between events.
Log Management
Ingest and retain 100% of logs at petabyte scale. No sampling, no data loss, no compromises.
AI SRE
Network of domain-specific AI agents that investigate incidents, generate hypotheses, and accelerate resolution.
APM
OpenTelemetry-native application performance monitoring with full distributed tracing.
Kubernetes & Cloud
Real-time visibility into containers, pods, nodes, and cloud resources across any provider.
LLM Observability
Monitor AI model performance, token consumption, and agent behavior in production.
Service Level Management
Define SLOs, track error budgets, and get intelligent alerts before customers notice.
Funding Trajectory
From incubation to acquisition in seven rounds - $665M+ total raised
The Proof
Numbers don't lie. By 2024, Observe had achieved 200% year-over-year ARR growth and a 190% net revenue retention rate. That second number is remarkable: it means existing customers were nearly doubling their spend each year. They weren't just satisfied - they were expanding.
The company processes over 1 petabyte per day for individual enterprise customers, with sub-second query latency. That's not a benchmark demo. That's production workloads at some of America's largest financial institutions.
Trusted By
Observability is fundamentally a data problem, and Observe joining Snowflake is a natural extension of their AI Data Cloud.- Jeremy Burton, CEO, Observe
The Mission
Observe exists to eliminate the impossible trade-offs that have plagued IT operations for decades. Store everything or go broke. Query fast or query accurately. Hire more engineers or work them to exhaustion.
The mission is deceptively simple: turn business data into information. Make the signals actionable. Let machines do what machines are good at - correlating patterns across billions of events - so humans can do what they're good at: making decisions, building systems, solving problems that matter.
Why It Matters Tomorrow
When Snowflake announced the acquisition in January 2026, CEO Sridhar Ramaswamy called it an expansion into the $50+ billion IT operations management market. That's not hyperbole. As enterprises move to cloud-native architectures and deploy AI workloads at scale, the volume of telemetry data will only increase.
Observe's bet - that observability is fundamentally a data problem, not a tooling problem - has been validated in the most concrete way possible. A company that didn't exist a decade ago is now the observability layer for one of the world's most important data platforms.
The Scene, Revisited
It's still 3 AM in that operations center. But the engineer's screen looks different now. Instead of a dozen terminal windows and frantic Slack messages, there's a single dashboard. The AI SRE has already identified the problem, correlated it with a deployment from earlier that evening, and drafted a rollback plan.
The fix takes three clicks. The payment queue clears. Customers never notice.
This is what Observe built. Not a faster way to search logs - a fundamentally different relationship between engineers and their systems. One where the data works for you, instead of the other way around.
From a $1,500 contract with a regional telecom to a billion-dollar acquisition by the data cloud giant - in six years. That's not a pivot story or a growth hack story. It's what happens when you bet correctly on where the world is going, and then execute relentlessly until you get there.